Python joinpath Exemples

Exemple #1

for i in range(num_chains):
    with open(sampler_output_run_paths[i].joinpath('runtime.txt'), 'r') as file:
        runtimes.append(float(file.readline().rstrip()))
runtimes = np.array(runtimes)

# %% Drop burn-in samples

chain_arrays.vals['sample'] = chain_arrays.vals['sample'][:, diagnostic_iter_thres:, :]

# %% Compute multivariate rhat

rhat_val, _, _, _, _ = chain_arrays.multi_rhat(mc_cov_mat=mc_cov_mats)

# %% Save rhat_val

with open(sampler_output_path.joinpath('multi_rhat.txt'), 'w') as file:
    file.write('{}\n'.format(rhat_val))

# %% Compute multivariate ESS

ess_vals = np.array(chain_arrays.multi_ess(mc_cov_mat=mc_cov_mats))

# %% Save multivariate ESSs

for i in range(num_chains):
    with open(sampler_output_run_paths[i].joinpath('multi_ess.txt'), 'w') as file:
        file.write('{}\n'.format(ess_vals[i]))

# %% Save mean of multivariate ESSs

with open(sampler_output_path.joinpath('mean_multi_ess.txt'), 'w') as file:

Exemple #2

Fichier : benchmark_pred_accuracy_via_mean.py Projet : papamarkou/bnn_mcmc_examples

# %% Load packages

import numpy as np
import torch

from sklearn.metrics import accuracy_score

from bnn_mcmc_examples.examples.mlp.penguins.constants import num_chains
from bnn_mcmc_examples.examples.mlp.penguins.dataloaders import test_dataloader
from bnn_mcmc_examples.examples.mlp.penguins.metropolis_hastings.constants import (
    sampler_output_path, sampler_output_run_paths
)

# %% Load test data and labels

_, test_labels = next(iter(test_dataloader))

# %% Compute predictive accuracies

accuracies = np.empty(num_chains)

for i in range(num_chains):
    test_preds = np.loadtxt(sampler_output_run_paths[i].joinpath('preds_via_mean.txt'), skiprows=0)

    accuracies[i] = accuracy_score(test_preds, torch.argmax(test_labels, 1))

# %% Save predictive accuracies

np.savetxt(sampler_output_path.joinpath('accuracies_via_mean.txt'), accuracies)

Exemple #3

import numpy as np

import kanga.plots as ps

from bnn_mcmc_examples.examples.mlp.penguins.constants import num_chains
from bnn_mcmc_examples.examples.mlp.penguins.metropolis_hastings.constants import (
    sampler_output_path, sampler_output_run_paths)

# %% Load correlation matrices

mc_cor_mats = []
for i in range(num_chains):
    mc_cor_mats.append(
        np.loadtxt(sampler_output_run_paths[i].joinpath('mc_cor.csv'),
                   delimiter=',',
                   skiprows=0))
mc_cor_mats = np.stack(mc_cor_mats)

mean_mc_cor_mat = np.loadtxt(sampler_output_path.joinpath('mean_mc_cor.csv'),
                             delimiter=',',
                             skiprows=0)

# %% Plot heat maps of correlation matrices

for i in range(num_chains):
    ps.cor_heatmap(mc_cor_mats[i],
                   fname=sampler_output_run_paths[i].joinpath('mc_cor.png'))

ps.cor_heatmap(mean_mc_cor_mat,
               fname=sampler_output_path.joinpath('mean_mc_cor.png'))

Exemple #4

Fichier : benchmark_mc_cov.py Projet : papamarkou/bnn_mcmc_examples

    'sample'][:, diagnostic_iter_thres:, :]

# %% Compute Monte Carlo covariance matrices

mc_cov_mats = chain_arrays.mc_cov()

# %% Save Monte Carlo covariance matrices

for i in range(num_chains):
    np.savetxt(sampler_output_run_paths[i].joinpath('mc_cov.csv'),
               mc_cov_mats[i],
               delimiter=',')

# %% Save mean of Monte Carlo covariance matrices

np.savetxt(sampler_output_path.joinpath('mean_mc_cov.csv'),
           mc_cov_mats.mean(0),
           delimiter=',')

# %% Compute Monte Carlo correlation matrices

mc_cor_mats = chain_arrays.mc_cor(mc_cov_mat=mc_cov_mats)

# %% Save Monte Carlo correlation matrices

for i in range(num_chains):
    np.savetxt(sampler_output_run_paths[i].joinpath('mc_cor.csv'),
               mc_cor_mats[i],
               delimiter=',')

# %% Save mean of Monte Carlo correlation matrices